Background
Chronic inflammatory arthritides, such as rheumatoid arthritis (RA) and psoriatic arthritis (PsA), lead to substantial changes in the articular cartilage. On the one hand there is net loss of articular cartilage since synovitis induces the release of factors that inhibit essential pathways for matrix synthesis by chondrocytes and trigger catabolic pathways such as the expression of matrix-degrading enzymes [
1]. Interleukin-1 (IL-1) and tumour necrosis factor alpha (TNFα), for instance, are potent cartilage-degrading factors released from the synovial membrane, which induce chondrocyte dedifferentiation and matrix breakdown. The latter is guided by the excessive release of matrix-degrading enzymes such as members of the A disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) and matrix metalloproteinase (MMP) families [
2‐
4].
On the other hand, arthritis can also lead to local cartilaginous proliferations. In this case, osteo-chondroprogenitor cells in the periosteum start to proliferate and form condensations. These structures then undergo chondrogenic differentiation with deposition of cartilage matrix and form a template for ossification. The innermost chondrocytes mature towards hypertrophic chondrocytes, secreting a matrix rich in type X collagen, which becomes mineralised and eventually replaced by bone [
5]. In this process, hypertrophic chondrocytes may also transdifferentiate into osteoblasts [
6].
Apart from external factors such as cytokines, which affect cartilage homeostasis during arthritis, we hypothesised that there are factors which are produced by the cartilage itself and regulate cartilage homeostasis during arthritis. We previously identified the secreted Upper zone of growth plate and cartilage matrix associated protein (Ucma; also termed Gla-rich protein (GRP)) in a screen for cartilage-specific genes [
7]. Ucma expression in mice is confined to cartilage, although it also migrates through the matrix and gets released from the cartilage compartment [
7‐
10]. Knock-down of Ucma in zebrafish results in cartilage defects and a severely disturbed craniofacial development [
11], while Ucma-deficient mice developed normally, without any overt alterations in cartilage and skeletal development [
9]. Ucma-deficient mice, however, develop more severe experimental osteoarthritis after destabilisation of the medial meniscus (DMM), indicating a chondroprotective effect of Ucma [
8].
Although of different etiopathogenesis, the terminal pathways leading to cartilage degeneration in osteoarthritis and inflammatory arthritis may be similar. In this study, we therefore investigated the nature of Ucma-dependent aggrecanase inhibition and the effect of Ucma on cartilage homeostasis during inflammatory arthritis. In our experiments we show that Ucma directly inhibits aggrecanase activity by physical interaction with ADAMTS5. Moreover, we demonstrate that Ucma-deficient mice developed more severe arthritis-triggered cartilage degradation, while they did not differ from wild-type mice in terms of clinical signs of arthritis and the extent of synovial inflammation. On the other hand, treatment with recombinant Ucma ameliorated cartilage degeneration during arthritis, suggesting that Ucma has protective effects on the cartilage. Ucma was not only expressed in articular cartilage but was also found in osteophytes. Ucma deficiency also affected the development of osteophytes with fewer osteophytes, osteoblasts and osteoclasts in Ucma-deficient mice during serum-induced arthritis, reflecting Ucma-dependent regulation of osteoblasts and/or osteoclasts, which are required for osteophyte formation and growth.
Discussion
This study identifies the cartilage-derived protein Ucma as a protective factor for cartilage loss in the context of inflammatory arthritis. Ucma physically binds to ADAMTS5 and inhibits its aggrecanase activity. Consequently, cartilage loss during arthritis is significantly more severe in Ucma-deficient mice than in WT mice but is alleviated when arthritic mice are treated with recombinant Ucma.
Aggrecanases of the ADAMTS family play a prominent role in cartilage destruction during pathological joint degeneration. They are considered to mediate one of the first steps in cartilage degradation by cleaving aggrecan within its interglobular domain, resulting in the release of the aggrecan chondroitin sulphate (CS)-rich region into the synovial fluid. This directly compromises joint function. Moreover, after the loss of the aggrecan CS-rich region, denuded collagen fibrils are more susceptible to collagenase cleavage and thereby ADAMTS aggrecanases contribute to further irreversible steps of cartilage degradation [
3]. ADAMTS5 appears to be the primary aggrecanase, which is responsible for aggrecanolysis and cartilage degeneration in murine experimental osteoarthritis, while evidence rises for a contribution of both ADAMTS5 and ADAMTS4 in human cartilage degradation [
2,
18‐
20].
Due to the pivotal role of these aggrecanases in cartilage degradation, their inhibition during joint pathologies is considered to be a primary aim in the treatment of cartilage degeneration. At least in murine models of arthritis, anti-aggrecanase regimens have provided promising results. Thus, intra-articular administration of monoclonal antibodies against ADAMTS5 has been shown to effectively ameliorate disease progression in a spontaneous mouse model of OA [
21]. Other effective aggrecanase inhibitors have been shown to exert their inhibitory potential by physically biding to the thrombospondin type-1 repeats of ADAMTS4 and ADAMTS5. Thus, a truncated form of tissue inhibitor of metalloproteinases 3 (N-TIMP3) interacts with ADAMTS4 and ADAMTS5, thereby inhibiting their aggrecanase activity [
22,
23]. Synthetic inhibitors of aggrecanases have also been designed to bind to ADAMTS4 and/or ADAMTS5 to inhibit their aggrecanase activity [
3].
We have recently demonstrated that Ucma inhibits ADAMTS4 and ADAMTS5 aggrecanase activity in vitro and that Ucma-deficient mice develop a more severe cartilage damage in experimental osteoarthritis [
8]. Here, we demonstrate that Ucma physically interacts with ADAMTS5. Moreover, recombinant Ucma inhibited ADAMTS5-triggered NITEGE formation in cultured 4C6 chondrocytes. These findings indicate that Ucma-dependent aggrecanase inhibition is directly mediated by physical interaction of Ucma with ADAMTS aggrecanases.
The pivotal role of ADAMTS aggrecanases in cartilage degradation is not confined to degenerative joint diseases such as osteoarthritis. ADAMTS aggrecanases are also key factors for cartilage degradation in inflammatory arthritis. Thus, Stanton et al. [
2] have demonstrated that ADAMTS5-deficient mice are protected from cartilage degeneration during antigen-induced arthritis.
Consistently, our findings suggest that Ucma also protects from cartilage loss during inflammatory arthritis. The pro-inflammatory and arthritis-associated cytokine IL-1β is a potent inducer of ADAMTS activity in chondrocytes [
24]. In line with this notion, IL-1β induced the formation of ADAMTS-specific aggrecan cleavage products (NITEGE neoepitope) in cultured 4C6 chondrocytes. Recombinant Ucma inhibited this IL-1β-triggered formation of NITEGE neoepitopes in 4C6 cultures, indicating that Ucma inhibits aggrecanase activity in cartilage under inflammatory conditions. Moreover, although
Adamts4 and
Adamts5 mRNA levels were not altered in Ucma-deficient mice, NITEGE neoepitope formation and cartilage damage were found to be significantly increased in articular cartilage of Ucma-deficient mice with SIA. Thus, these findings are consistent with the notion that Ucma is an inhibitor of ADAMTS-specific proteolytic cleavage of aggrecan also in inflammatory arthritis. Consequently, systemic administration of Ucma protected mice from cartilage degeneration during SIA.
These data are remarkable in the light that inflammation was no different between WT and Ucma-deficient mice or WT mice after Ucma treatment, which excludes indirect effects of Ucma on cartilage through potential down-regulation of inflammation. Synovial inflammation and clinical signs of arthritis were indistinguishable between WT and Ucma-deficient mice. Previous data by Cavaco et al. [
25] indicated a possible anti-inflammatory potential of Ucma. Thus, recombinant Ucma reduced IL-1-induced gene expression of cyclooxygenase 2 (COX2) in human primary chondrocytes and synovial fibroblasts. Consequently, IL-1β-induced prostaglandin E2 (PGE2) secretion was decreased upon Ucma treatment [
25]. The same group also observed that recombinant Ucma reduced TNFα and PGE2 production of LPS-stimulated macrophage-like cells [
26]. In vivo, however, these effects seem to be of minor importance as the activity of SIA, which depends on cytokines such as IL-1 and TNFα, was not altered in the absence of Ucma or after systemic administration of recombinant Ucma.
While Ucma expression is strictly confined to the articular cartilage in steady-state conditions, additional expression of Ucma was also found in growing osteophytes in SIA, implying a possible role of Ucma also in SIA-related osteophyte formation or growth. In fact, SIA-triggered osteophyte formation was reduced in Ucma-deficient mice. Osteophyte formation usually resembles the process of endochondral ossification, where a cartilage primordium gets replaced by bone via an intermediate stage of calcified cartilage tissue rich in collagen type X [
5]. Osteophytes in SIA are also formed via a cartilage primordium, which is reflected by the presence of
collagen10a1 expressing hypertrophic chondrocytes in developing osteophytes. Interestingly, however, we did not observe any Ucma-dependent difference in
collagen10a1-positive hypertrophic chondrocytes at sites of osteophyte formation.
his finding may indicate that the formation and maturation of the cartilaginous osteophyte primordium is not affected by Ucma. The size of osteophytes, however, is not only determined by the modelling of the cartilage primordiae but also by its remodelling by osteoblasts and osteoclasts [
5]. Intriguingly, reduced osteophyte formation in Ucma-deficient mice was associated with a decrease in osteoclasts and osteoblasts. Like in the case of experimental osteoarthritis [
8], Ucma-dependent differences in osteoclast numbers were not associated with changes in
Rankl/Opg mRNA ratios. This supports our earlier hypothesis that Ucma can stimulate osteoclast differentiation in a RANKL-independent manner. In fact, we demonstrated previously that Ucma can stimulate osteoclastogenesis in vitro and stimulation of pre-osteoclasts with Ucma rapidly induces the phosphorylation of MAP kinases, which may mediate such a RANKL-independent pro-osteoclastic effect [
8,
27,
28]. Consistent with the observed decrease in osteoblast numbers in Ucma-deficient mice during SIA, Lee et al. [
29] have previously reported that overexpression of Ucma in pre-osteoblasts promoted osteogenic differentiation in vitro. These data indicate that Ucma affects osteophyte formation on the level of bone remodelling rather than via regulating cartilage maturation.
While systemic administration of recombinant Ucma affected cartilage damage it did not elicit significant effects on osteophyte formation. This finding, however, may be explained by the earlier time point of histological analysis. Ucma-deficient mice and WT littermates have been histologically examined at 14 days after serum transfer, when osteophytes were fully developed. After Ucma administration the mice were analysed already 10 days after serum transfer, when osteophytes were still developing. This earlier time point was chosen for optimal detection of differences in cartilage degradation. However, as we have demonstrated, early stages of osteophyte development (formation of cartilage primordiae) do not appear to be affected by Ucma. Yet there may also be differences in the activities and local concentrations of endogenous and recombinant Ucma, which might also account for this, at first sight, unexpected result.